The Future of Legal Artificial Intelligence
Artificial intelligence is the use of machine learning to allow computers to analyze data, learn patterns, and produce insights from the data. Artificial intelligence applications for legal services include review of documents, conducting legal research, performing due diligence, contract management, and even predicting legal outcomes.[i] A few applications of legal AI are already commercially available, and the field is growing. The bulk of this article is focused on ROSS Intelligence, Inc., whose legal research platform will transform how lawyers work very soon.
ROSS Intelligence, Inc.
ROSS is a legal research platform powered by artificial intelligence for U.S. law.[ii] It uses its AI systems to solve two problems lawyers run into when performing legal research: (1) too many cases on a legal issue or (2) too few cases on a legal issue. ROSS’ search process breaks down the research process into understanding the legal issue, retrieving relevant resources on the issue, and ranking the results. ROSS does not just rank results by number of keyword hits, but uses machine learning, grammatical structure, word embeddings, and “facts and motions” to rank relevant information.
ROSS’s AI platform transforms the legal research in each of these steps.[iii] Take the example question, “What is the standard for gross negligence in Oklahoma after 2004?”
Machine learning is the bedrock of AI systems. It is a way to teach computes to learn from themselves by recognizing patterns of context, syntax, and meaning within legal documents. That is to say, the more ROSS does, the better it gets (hopefully true of your associates as well).
Next is grammatical structure. ROSS understands grammatical structure and the way the meaning of sentences changes depending on the relationships between types of words. In this example, ROSS knows that gross is an adjective that modifies negligence and thus groups the terms together when doing its research. Also, the same keywords do not yield the same results. ROSS knows that “the boy loves a girl” has a different meaning from “the girl loves a boy,” although they have the same words (it also understands that it is similar to “the man loves a woman”). It breaks all of this down without Boolean search modifiers like quote marks, asterisks, and indicating the proximity of words.
Word embedding is a mathematical technique for placing similar words together. This allows ROSS to find words that are similar to or related to the words in a search and compare them to those in cases. In our example question, lawyers come to understand that “duty” and “negligence” are related. Ross uses that relationship to broaden its understanding beyond keywords (and potentially return results faster and better than your first-year associates).
All of this would be great alone, but the next step may be what will set ROSS as the standard going forward. Ross is designed to consider additional context including the facts of the case, procedural posture, and type of motion. ROSS’s results highlight the additional context as it is identified and thus help you identify more relevant cases.
The platform can also analyze a legal document (a motion or brief you receive or draft), and return a comprehensive list of cases that have received negative treatment. It can also store queries, monitor updates in the law, and send notifications to users with relevant legal updates.
Up and Coming Applications: Contract Review & Automated Divorce
LawGeex uses AI to simplify contract review.[iv] It automatically reviews every contract that enters your system, understands the legal language and business implications (and compares them to the users set policies), flags missing or non-compliant clauses (and offers one-click correction), and prepares the approved or corrected contract for export. j
Other major players in contract review include Kira[v] and eBrivia.[vi]
WeVorce is an online uncontested divorce platform.[vii] It guides users through its three-step process of (1) defining optimal outcomes, (2) creation of court-ready documents, and (3) delivery of documents and filing instructions. Its system creates custom for each State. It can produce Marriage Settlement Agreements, Parenting and child-related agreements, Child financial support documents, Name change provisions, Spousal Support Plans, and more.
Impact of Legal AI
Reports predict that 100,000 legal roles will be automated by 2036.[viii] As technology such as artificial intelligence proliferates, the legal services industry will have to decide how to adapt. The market will surely implement these advances even if traditional law firms do not (e.g. LegalZoom).
Artificial intelligence has the capability to not only change how law firms work, but change how legal services are delivered all together. It does not take much imagination to see a future where consumers can access relevant and detailed legal information directly from AI platforms without going through attorneys. If we want to shape the outcome of AI development, it is crucial that we engage now. I think the future is sooner than we think.